SMPS with output adjustable from 0-100% - what topology?

I need a power supply with an output adjustable from 0 to 12Vcc-10A. I'm not sure if I make a 12Vcc fixed supply and then put it trought a PWM switch with a filter to adjust the amplitude, or I try to make a supply with 110v/220vac input to 0-12Vcc (flyback or forward).

I am afraid a single SMPS won't be able to generate an output which I can adjust from 0 to 100%.

• The topology is defined by the use case, and then parts are chosen to match. – Ignacio Vazquez-Abrams Dec 23 '16 at 22:10
• Your requirements are unclear and messy. Making assumptions like I am afraid a single SMPS won't be able to generate an output which I can adjust from 0 to 100℅ makes this question confusing. No one specifies SMPS output in %, they have an output voltage range. There are SMPSs which can regulate their output down to 0 V. Look up specifications for some SMPSs and write your requirements in a similar format. The topology cannot be discussed until you have your requirements in order. To choose a topology you most know pros and cons of all, which you don't. – Bimpelrekkie Dec 23 '16 at 22:14

First, for 10A output current, flyback would be totally the wrong choice due to high losses on output windings and high flyback spikes to the primary. Two-switch flyback may be a good choice but still the output winding is the issue for 10A current. Go for either two-switch forward, half bridge or even LLC half bridge.

Designing a variable output SMPS is a challenging task:

The first problem comes from generating supply voltage for the controller block: The auxiliary winding concept in conventional isolated converters is totally useless for variable SMPS, because the output voltage of this winding will be load- or output voltage-dependent. So you'll need a separate supply. Actually, a self-oscillating (also known as "Ringing Choke") converter with, say, 15V/200mA output will do the job. But, as you might guess, cost and placement issues... Still it's the best solution.

The another problem is generating 0V. IIRC, some articles about this tells that a negative reference is needed. But I'm not sure. Anyway, if your minimum output voltage requirement was 2.5VDC then simple TL431 and divider network with a linear pot would simplify the design.

And the last problem is light load or even unloaded state. For forward and half-bridge, you need an output inductor. It's easy to get 10A output from a 33uH toroidal inductor with a switching frequency of 100kHz. But what if the output current goes down to 10mA (Suppose you want to test an LED with your supply, for example)? You'll need either an output inductor with high inductance (say, 1mH) or a heavy dummy load to waste some of output power for unloaded state which decreases efficiency. LLC half-bridge does not need an output inductor but even for this, you still need to waste at least %5 of max output power on a dummy load. Of course you can get the converter working with 12VDC/10mA output even with %1 dummy load, but if you decrease the output voltage then the converter will enter discontinuous mode and you'll see that the output voltage is fluctuating.

Rather than output voltage, you'll also need to regulate output current (for current limiting or output shorting). This will need an extra comparator which brings extra care for designing output and feedback blocks (feedback stabilization).

• Thanks Rohat for the detailed answer. So, I think it may be simpler to make a 12V-10A fixed supply and then put it through a PWM switch to generate the variable voltage, and use a LC circuit to filter it. – Alexandre Bueno Dec 24 '16 at 10:43
• So you'll put a P-Ch MOSFET on the (+) line and a LC filter, right? Note that the load current fill flow through the inductor, so you'll need a large one. And cut-off frequency of the filter should be quite low (about one tenth of PWM freq.), so you'll need either a large inductor or a large cap. The latter won't allow the output voltage to decrease or increase quickly. Another thing to take into consideration is spikes and ringing due to switching. And the last thing: You'll also switch -so reduce- the load current and the converter won't allow you to draw 10A when the output voltage is low. – Rohat Kılıç Dec 24 '16 at 14:41
• Yes, I have designed a LC filter with a MOSFET in series with the Vdd. I have also put a diode from the GND to the MOSFET drain, so there will be a way to the inductor's flow when the switch is off. Then I realized it was exactly a buck converter. But it's ok. I have no need to increase or decrease the output quickly. And also, when the voltage is low, the current will lower too, the charge is like a resistor. – Alexandre Bueno Dec 25 '16 at 11:31